Date of Award
Doctor of Philosophy
Electrical and Computer Engineering
Dr. Ikhlas Abdel-Qader
Dr. Johnson Asumadu
Dr. James Springstead
Mass Spectrometry (MS) data is ideal for identifying unique bio-signatures of diseases. However, the high dimensionality of MS data hinders any promising MS-based proteomics development. The goal of this dissertation is to develop an accurate classification tool by employing compressive sensing (CS). Not only can CS significantly reduce MS data dimensionality, but it also will allow for full reconstruction of original data. The framework developed in this work is based on using L2 and a mixed L2-L1 norms, allowing an overdetermined system to be resolved. The results show that the L2- based algorithm with regularization terms has a better performance than that of L1 and Q5 algorithms under all applicable assumptions. Performance was measured using overall success rate, sensitivity, positive predictive value and specificity. The regularization parameters and sensing matrix were optimized to achieve a robust classification method. Additionally, the Block Sparse Bayesian Learning (BSBL) algorithm was used to reconstruct MS data using a fingerprint-based technique. The simulation results validate the proposed framework and indicate the potential for a successful prostate cancer detection technique using MS data. The proposed framework can be a useful tool for assessing patient risk of disease and will aid in paving the way for personalized medicine.
Restricted to Campus until
Awedat, Khalfalla Ahmad Kh., "Compressive Sensing Framework for Mass Spectrometry Data Analysis" (2016). Dissertations. 1429.